@inproceedings{liu-etal-2026-retrieval,
title = "Retrieval-{GRPO}: Reinforcement Learning for Dense Retrieval in Taobao Search",
author = "Liu, Xingxian and
Li, Dongshuai and
Wan, Jiahui and
Wen, Tao and
Ling, Gui and
Lv, Fuyu and
Yan, Yuliang and
Ou, Dan and
Tanghaihong and
Zheng, Bo",
editor = "Atanasova, Pepa and
Haffari, Gholamreza and
Wang, Minghan",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-industry.73/",
pages = "1234--1245",
ISBN = "979-8-89176-460-6",
abstract = "Dense retrieval is a fundamental component of e-commerce search, projecting user queries and items into a shared semantic representation space for large-scale, real-time retrieval. While LLM-based embedding models are increasingly replacing BERT-style encoders, prevalent training pipelines still largely rely on BERT-like supervised fine-tuning and hard negative mining. Such paradigms require intricate offline negative construction, thereby constraining model iteration efficiency and limiting the evolution of learned semantic representations. We propose Retrieval-GRPO, a dense retrieval framework based on reinforcement learning. It dispenses with offline hard negative mining by performing dynamic Top-K candidates retrieval during training. Moreover, we introduce a relevance-oriented LLM as a reward model to provide real-time supervisory signals. Guided by the resulting reinforcement learning objective, the retriever updates embeddings to better capture semantic correspondence and achieve improved generalization on complex long-tail queries. Extensive offline and online experiments substantiate the effectiveness of Retrieval-GRPO in Taobao Search, which serves China{'}s largest e-commerce platform."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-retrieval">
<titleInfo>
<title>Retrieval-GRPO: Reinforcement Learning for Dense Retrieval in Taobao Search</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xingxian</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dongshuai</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiahui</namePart>
<namePart type="family">Wan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tao</namePart>
<namePart type="family">Wen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gui</namePart>
<namePart type="family">Ling</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fuyu</namePart>
<namePart type="family">Lv</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuliang</namePart>
<namePart type="family">Yan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dan</namePart>
<namePart type="family">Ou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name>
<namePart>Tanghaihong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gholamreza</namePart>
<namePart type="family">Haffari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-460-6</identifier>
</relatedItem>
<abstract>Dense retrieval is a fundamental component of e-commerce search, projecting user queries and items into a shared semantic representation space for large-scale, real-time retrieval. While LLM-based embedding models are increasingly replacing BERT-style encoders, prevalent training pipelines still largely rely on BERT-like supervised fine-tuning and hard negative mining. Such paradigms require intricate offline negative construction, thereby constraining model iteration efficiency and limiting the evolution of learned semantic representations. We propose Retrieval-GRPO, a dense retrieval framework based on reinforcement learning. It dispenses with offline hard negative mining by performing dynamic Top-K candidates retrieval during training. Moreover, we introduce a relevance-oriented LLM as a reward model to provide real-time supervisory signals. Guided by the resulting reinforcement learning objective, the retriever updates embeddings to better capture semantic correspondence and achieve improved generalization on complex long-tail queries. Extensive offline and online experiments substantiate the effectiveness of Retrieval-GRPO in Taobao Search, which serves China’s largest e-commerce platform.</abstract>
<identifier type="citekey">liu-etal-2026-retrieval</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-industry.73/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>1234</start>
<end>1245</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Retrieval-GRPO: Reinforcement Learning for Dense Retrieval in Taobao Search
%A Liu, Xingxian
%A Li, Dongshuai
%A Wan, Jiahui
%A Wen, Tao
%A Ling, Gui
%A Lv, Fuyu
%A Yan, Yuliang
%A Ou, Dan
%A Zheng, Bo
%Y Atanasova, Pepa
%Y Haffari, Gholamreza
%Y Wang, Minghan
%A Tanghaihong
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-460-6
%F liu-etal-2026-retrieval
%X Dense retrieval is a fundamental component of e-commerce search, projecting user queries and items into a shared semantic representation space for large-scale, real-time retrieval. While LLM-based embedding models are increasingly replacing BERT-style encoders, prevalent training pipelines still largely rely on BERT-like supervised fine-tuning and hard negative mining. Such paradigms require intricate offline negative construction, thereby constraining model iteration efficiency and limiting the evolution of learned semantic representations. We propose Retrieval-GRPO, a dense retrieval framework based on reinforcement learning. It dispenses with offline hard negative mining by performing dynamic Top-K candidates retrieval during training. Moreover, we introduce a relevance-oriented LLM as a reward model to provide real-time supervisory signals. Guided by the resulting reinforcement learning objective, the retriever updates embeddings to better capture semantic correspondence and achieve improved generalization on complex long-tail queries. Extensive offline and online experiments substantiate the effectiveness of Retrieval-GRPO in Taobao Search, which serves China’s largest e-commerce platform.
%U https://aclanthology.org/2026.emnlp-industry.73/
%P 1234-1245
Markdown (Informal)
[Retrieval-GRPO: Reinforcement Learning for Dense Retrieval in Taobao Search](https://aclanthology.org/2026.emnlp-industry.73/) (Liu et al., EMNLP 2026)
ACL
- Xingxian Liu, Dongshuai Li, Jiahui Wan, Tao Wen, Gui Ling, Fuyu Lv, Yuliang Yan, Dan Ou, Tanghaihong, and Bo Zheng. 2026. Retrieval-GRPO: Reinforcement Learning for Dense Retrieval in Taobao Search. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 1234–1245, Budapest, Hungary. Association for Computational Linguistics.